The KillerDAC Audio forum
General HIFI => Other DACs => Topic started by: stevenvalve on November 25, 2017, 01:50:22 AM
-
Ian was over a few days ago with his full house modified Marantz 94 Mark2. He has spent many years perfecting it to this level. In its present state it is just brilliant and yes it has no valves. Its the only DAC that has been through this door that went neck and neck with my killerDac, it was very clean clear and noise free (one of his aims) Black indeed around the notes and they appeared to come from empty space. The sound stage is not as big and it does not have quite the bloom as the KillerDac, but is very similar in timbre and texture as mine and what i really like is it is not bright or white, instruments sound real, unlike most digital playback. It has none of the slow plodding nature of standard Marantz 94s. It is relaxed, natural, smooth, and has presents, with a solid image focus, yum its nice. Shows you can build brilliant from just good. But before you run out and buy a Marantz 94 remember he has spent many years tuning this machine and he knows how to get all the hard to find esoteric capacitors and resistors, and most are not commercially available. Anyway i will not rave on any more, but hopefully he can come to this page and explain how it works. The pictures do not show the full story, because underneath there is a compartment stuffed with more toys. I really could live with this Marantz because it makes Music, They should have built them like this in the first place.
-
Hi Steve and Ian :)
Yes indeed it would be great hearing a blow by blow account of the upgrade path Ian has traveled in order to be able to be praised by SG!!! It must have a couple of great sounding TDA chips inside it then. What other equipment is the CD spinner being used with??
For something with OP-amps in the output chain it does not always mean it's bad............I recently put some dexa discrites in place of some OP-amps and was shocked at how much better it performed all round. I tired to avoid the more usual Burson's etc.
It is easy to spot a lot of the changes implemented, the changed dac decoupling caps, the obvious clock board and the diode replacements. I remember those original POOGE articles and that couple of hundred page Marantz CD65 upgrades on the DIY hifi site.
I would love for Ian to describe the process and give a guide to the uplift various stages took on. This website could do with more helpful information and it will bring increased readership I feel. ;D ;D ;D ;D
-
Hi guys . I am travelling back home to Newcastle at the moment. Rest assured I will post a detailed account of my experiences with the Marantz cd 94 mk 2 as well as other equipment upgrades.
One thing I will say now is that it is a holistic approach rather than using a super TDA1541 double crown chip for example . You need to reduce noise in every circuit because its all connected .
In digital , added noise means jitter . Capacitor quality both in Electrolytic and film bypasses are critical to take things to the next level . Attention to clock power supply is critical as well as the decoupling caps around the TDA1541 ( super sensitive ) .
Thanks again to Steve and Jen for being so welcoming .It was an absolute pleasure to spend quality audio time at their home .
-
Its difficult to know where to start. I could seriously write a book about my experiences .
i bought myself a brand new Marantz cd 94 mk2 around 1990 , but i was never satisfied with my sound and , after being on the merry go round for a number of years buying and borrowing high end equipment it seemed that every substitution always had tradeoffs when it came to musical satisfaction.
My gear was mostly excellent design and build quality such as the Stax mono blocs Classe DR6 so I came to the conclusion that it would be worth modifying my gear to add refinement to the reasonably dynamic sound i had .
During the late 90s i had performed literally thousands of mods and built up a memory bank of practical experience in what worked and what didnt . Not knowing aĺl that much in theory I wasnt constrained by preconceived ideas so I tried lots of radical stuff back then . I was even running battery clocks a decade before i realised there were other nutters out there 😊 . Eventually i achieved excellent sound quality through perseverance and trying to understand why things changed . For example this led me to studying capacitor construction quite intently .
Not algebraic theory mind you , rather , what different construction and materials bought to the table in a practical sense .
Anyway , kids came along and the system was totally dismantled sold or stored .
A few years ago i decided to rebuild a sound system based around the 94 and was confident i could improve on my original.
The Marantz cd 94 machines had all the ingredients for greatness . Arguably the best drive ever made in the CDM 1 and possibly the best chip ever made in the tda1541a ,excellent multiple regulated power supplies with generally good parts quality for a commercial product.
So what do we do to improve it if its so good ? Well the devil is in the details.
Firstly , to achieve greatness ,a component is only as good as its weakest link. There is no single place to start really .Its a matter of identifying and working through every single weak link till they are all eliminated or improved .
In the 90s I did a shootout with probably 20 different op amps .The bottom line without boring everyone about the differences was that the AD 827 and AD 847 were the clear favourites being excellent in all areas including dynamics which seemed to be a problem for many op amps being too smooth or bland . The originals in the 94 also had good dynamics but are coarse or ragged in comparison.
These are the op amps in my machine today . Not cheap but very capable. Op amps upgrade probably gave a few percent of the improvement in sound quality so dont expect miracles from any single mod or you will be sadly disappointed as with any of the other mods in isolation .As I said it is a wholistic approach with each mod being rather small incremental improvements
I will detail some more of the 94 changes in a number of posts as there is a fair bit to get through ,so i will end this one for now .. Cheers Ian .
-
Well, I for one will look forward to your audio travels IRT the Marantz silver spinner. Of course having two of them I have an obvious interest on finding the things that worked ;D
-
Time to buy a Marantz 94. Some say the Marantz 94 with a single chip was a better idea than the mark 2, with 2 Chips. They feel going to the two chips compromised the performance in some ways. I have no idea, any thoughts. Question is buy a mark 1 or 2.
-
Great question Steve ,and one i have wrestled wit a lot . I owned both at the same time and compared them side by side . Firstly the mk2 had a bit more weight . It probably had more drive due to the doubling up of dacs and amplification stages in dual differential .
The mk1 was a bit sweeter and and maybe a bit less grungy . This was almost certainly due to the better film caps around the tda1541 a even though they werent great caps ( mylar , inductive)
Certainly not optimum . As you could hear ( comparatively ) grungy plodding in the mk2 can be totally fixed .
I am not 100 pc sure the weight of the mk1 can be made to equal the mk2 but certainly could be improved .
I think the dual differential architecture is superior as far as noise and distortion are concerned but the extra room in the mk1 opens up lots of possible choices for even better exotic decoupling caps around the tda1541 and for top line caps the cost will be halved needing 14 instead of 28 .
The caps in the CD12 and CD 7 around the dac were top quality non inductive polypropylene foil types i believe . These would be impossible to fit on the tiny mk2 board .I was lucky enough to get a few of those green pps copper foils for mine , but they are no more The only ones I have ever seen are the ones i got .
The mounting arrangement for the dac board in the mk 2 is also not optimum either , using steel prongs . This will be the last thing I will do .My last one was hard wired with Discovery signature wire .really nice wire and made a clear difference.
-
Great question Steve ,and one i have wrestled wit a lot . I owned both at the same time and compared them side by side . Firstly the mk2 had a bit more weight . It probably had more drive due to the doubling up of dacs and amplification stages in dual differential .
The mk1 was a bit sweeter and and maybe a bit less grungy . This was almost certainly due to the better film caps around the tda1541 a even though they werent great caps ( mylar , inductive)
Certainly not optimum . As you could hear ( comparatively ) grungy plodding in the mk2 can be totally fixed .
I am not 100 pc sure the weight of the mk1 can be made to equal the mk2 but certainly could be improved .
I think the dual differential architecture is superior as far as noise and distortion are concerned but the extra room in the mk1 opens up lots of possible choices for even better exotic decoupling caps around the tda1541 and for top line caps the cost will be halved needing 14 instead of 28 .
The caps in the CD12 and CD 7 around the dac were top quality non inductive polypropylene foil types i believe . These would be impossible to fit on the tiny mk2 board .I was lucky enough to get a few of those green pps copper foils for mine , but they are no more The only ones I have ever seen are the ones i got .
The mounting arrangement for the dac board in the mk 2 is also not optimum either , using steel prongs . This will be the last thing I will do .My last one was hard wired with Discovery signature wire .really nice wire and made a clear difference.
WRT 94 MKII:
They run dual differential but still do L and R ch in same DAC. So one TDA does L+ R+, other does L- R-.
Would have been better to use 1 x TDA L channel +and- / other R channel +and- but as they say, it is what it is. :)
They also used 5 opamps / channel in the dual dif configuration which was fairly complex.
Maybe this was to allow for FET switched de-emphasis circuit. I'd be pulling that out.
A simple 3 opamp config with class A opamps is the way to go.
I never tried the AD827 / AD847, however I got great results with OPA627 biased into class A
I recommend biasing them into class A - I have never found an opamp that didn't sound better with class A bias.
But as you say it's all a balancing act. The opamp PS bypass makes a difference too. Some people use double
caps, ie; electro with film cap but I think this gives slight artificial sound. With the 827 / 847 you have to be a
little careful as they are very fast and can oscillate. But it will usually not sound right if this is happening.
I used to enjoy doing this sort of simple modding to SS gear, it can be a bit maddening at times - this opamp,
that cap, etc etc
Oh and it's also worth trying running the analog OP stage straight off batteries with no regulation.
This can sound really good - if you don't mind being a battery slave..... charge etc :-X :)
T
-
This is the clock with integrated power supply all caps replaced one by one and assessed . Mostly Silmic 2 super golds but Silmic 2 standards very good too. Bypassed with Jupiter copper foils and pps copper foil caps This lowered the subjective noise floor immensely . I also added a couple of film caps very close to the active devices underneath .
The Silmics are much better than the supplied panasonics and philips caps
Not shown is a larger cap in parallel with the power supply . I think from memory 4700 uf elna for audio . The two 1000uf caps are just not enough for the clock .Much better sound with the bigger
capacitance . The clock power supply needs to be as noise free as possible and very reactive .so dont skimp on .01 bypasses . Dont use mylar here .
What seems to be most important here is to get the power supply to the clock as noise free as possible .noise means jitter . Jitter means loss of fidelity .
Would love to try a Zen clock here one day Terry . ☺
Here is also a photo of the new Valab clock with original capacitors .
-
Here is a photo of the big cap beside the clock module held in place by a resin capacitor clamp on ths base of the 94 .
-
Here are the tda1541 chips with the Panasonic pps copper foil capacitors
-
Ian,
Nice work. Quick question - Are you running digital filter or 0 x OS?
T
-
Hi Zen , its Non over sampling .
-
Moving along . The player has been converted to non oversampling using 99.999 annealed silver wire in teflon tube. Also added a couple of copper foil pps .01 bypasses closer to the chip
Attached photo .
-
I have boxed in the Duelund output capacitors underneath the player . There isn't enough room to fit them inside . To access underneath I simply need to unscrew the box section .the caps are affixed to the inside of box with double sided tape with some insulation over the caps and on the cd base plate.
None of this can be seen unless your eyesight is level with the underneath of the player .
This will be improved once modding is finnished with better and shorter wire and neater insulation .
I cut a small slot in the baseplate for the wires so the base plate can be removed without unsoldering wires .
The wires tuck in neatly to the space at the end of the box once fitted .
The box was the lid part of an electronic box from Jaycar . Perfect for this application .
-
Here is a better view of the clock mounting .
I screwed two solid brass standoffs that had to be extended a bit to get the exact length .
The clock board is very strong and rigid so it sits suspended above the marantz pcb toward the front of the player . Some of these jobs are very finniky and time consuming but worth the effort .
It had to be just above the marantz pcb but leave enough room for the larger caps to fit under the lid so the standoffs had to be measured perfectly .
-
The cap to the left is actually a beautiful Elna 6800 uf 35v big Starget . Wonderful caps . This is in parallel with the two silmics on the clock . I used a resin capacitor clamp Again , a very tight fit .
The size and the quality of the capacitance is extremely important here . It makes a big difference over the standard clock . I ended up using Jupiter copper foil .01 for bypasses which sounded best there but auricaps would have been good too .
The clamp is a snug fit around the capacitor but not too tight so I can slide it out if I want to work on the clock without unsoldering .
-
Another very important upgrade is the 6800uf cap in front of the two main filter caps .
The cap Marantz used is a lower grade Elna cap with steel leads , but is used to filter the critical digital circuit . I used the 6800uf Elna Starget here too .they have a beautiful rich but clear sound and is bypassed underneath with a copper foil polystyrene .01uf. This is way better than the stock cap with a very noticable lift in sound quality. Again , had to do some mucking around to fit this much larger size cap into this spot including moving a resistor to the underneath of the player.
The two Jupiter copper foils are bypassing the main filter caps ,giving a rich ,natural liquid sound .Big improvement.
-
Vitavox picked up the diode changes in an earlier post . On this board I have only done one lot so far .There three lots of power supply diodes and i have done this to other machines before and will eventually get the treatment.. You are probably wondering why the huge diodes ?
Well ,they are hexfreds ultra fast soft recovery (with the soft recovery part being the important bit ) and the ones I have used from different manufacturers like Harris HFA and international rectifier types sound similar . The ones I used in my first machine way back were the hfa08tb60 from memory they were in a smaller package and the legs had to be bent heavily to get them in .So i gave the big ones a try and they sounded just as good but the legs , being much wider made it much easier and neater to fit .i cant recall which ones these are probably the hfa 15 or 25 way over rated 😊. I did compare to shotkey diode way back but preferred the hexfred .I thought the shotkey diodes lost somerhing on the top end sort of like a treble cut in comparison .Some people like them but i prefer the Hexfreds . These diode replacements will reduce hash and are another welcome improvement . Like i said before , its the cumulative gains that get you there .
-
I have boxed in the Duelund output capacitors underneath the player . There isn't enough room to fit them inside . To access underneath I simply need to unscrew the box section .the caps are affixed to the inside of box with double sided tape with some insulation over the caps and on the cd base plate.
None of this can be seen unless your eyesight is level with the underneath of the player .
This will be improved once modding is finnished with better and shorter wire and neater insulation .
I cut a small slot in the baseplate for the wires so the base plate can be removed without unsoldering wires .
The wires tuck in neatly to the space at the end of the box once fitted .
The box was the lid part of an electronic box from Jaycar . Perfect for this application .
Wow - you have done a lot of work. I wish we could have spoken earlier - read on.
With a 2 DAC / differential to single ended arrangement such as the 94MKII, the output circuit can be direct coupled with no caps required.
This is always the best way to go if possible - it saves money and no cap is better than any cap in signal path - even Duelunds.
However the circuit needs to be a proper DIF -> SE design and it needs a small trimpot to null any offset at OP. This can be done pretty simply with 3 opamps.
The standard circuit uses 5 opamps per channel and is pretty complex. This is mainly because it incorporates Fet switched de emphasis which you don't need.
The opamps should be biased into class A - but the right amount needs to be tweaked.
The standard regulators are either 317 or 78xx type which don't filter or block rubbish coming from other parts of the player above a few kHz very well.
Once you get to 100's of kHz they almost stop working all together. This is one reason why your PS cap changes make so much difference, they are
doing most of the filtering work of the regs.
For regulators I use very simple discrete series pass or shunt are even better. These regs work right up into the MHz range at filtering incoming rubbish.
They also have much lower noise than 317 / 78xx regs - but I'm not totally convinced the random noise is a big issue.
It's also best to use separate regs for R and L channel.
All this can be made on a relatively small ground plane covered board with room for caps, resistors and with opamp sockets.
The baseline performance is a long way above what the standard CD94 gives you.
Then it's a case of tweaking resistor and cap types, opamp types and amount of class A bias to get everything sounding best or to your taste.
T
-
Thanks for that Terry . Marantz used 7812 regs for + -12 v but they also use number of discrete regs as well for the lower voltage sections like the dacs .
I always assumed that the first op amp on the dac board was a dual differential current to voltage conversion ,with the small transistors at the end of the board switching in the discrete filter for de emphasis . Then on to the second dual differential op amp amplifying stage , and lastly on to the single op amp as a buffer , but could be wrong .
Anyway I had considered using the trim pot on the single op amp to null offset , but wanted a baseline .so the best capacitor available would be the best base line to compare direct coupled .
I may need some tips on the correct way to get the best results for biasing class A.
I know Steve would like me to leave it alone but there is still a bit to do to this player particularly things I have done in the past that i know will make it better .
One thing I did want to do is to showcase the real potential of Marantz's tda1541 , 94 mk2 and what it is capable of without changing the basic circuitry too much except for the clocking improvements .Even the NOS is not a major fundamental change to the original design .
I want to highlight that this eighties technology could cut it without resorting to major redesigning and also to highlight the importance of the supporting components as the major limiting factor in much audio equipment , not just CD players .
-
Hi Terry , this is the first op amp after the dac and the de emphasis circuit .you would understand it much better than me .
-
Thanks for that Terry . Marantz used 7812 regs for + -12 v but they also use number of discrete regs as well for the lower voltage sections like the dacs .
Yeah - I'm asleep, doh - Homer Simpson moment, haha! ;)
It does have discrete final regs for R and L channel and they look pretty good. So well done Marantz on that count.
I always assumed that the first op amp on the dac board was a dual differential current to voltage conversion ,with the small transistors at the end of the board switching in the discrete filter for de emphasis . Then on to the second dual differential op amp amplifying stage , and lastly on to the single op amp as a buffer , but could be wrong .
Anyway I had considered using the trim pot on the single op amp to null offset , but wanted a baseline .so the best capacitor available would be the best base line to compare direct coupled .
1st 2 OPA's are I-Vs for each phase with de emphasis. There is also some low pass filtering on this stage.
2nd 2 OPA's are for more low pass filtering (2nd order) These are C701/702, C703/704. Again each phase is separate.
3rd single OPA is the DIF to SE converter / OP driver.
I think they prolly use the 3 stages to get the 3 orders of low pass filtering. The low pass filtering removes noise from above audio band.
There are a lot of views WRT how much low pass filtering you need. Most of these early players had 3 orders of filtering. I think for
multibit DACs you don't need much.
With the Killer DAC we ended up using none at all. From memory Steve said it did some damage to the sound so I explained how to remove it.
The low pass filtering does have a pretty big effect on the sound - but with a SS opamp circuit (as opposed to valve) I think you still need some.
Tweaking the LPF on a circuit like the 94MK2 is pretty much beyond simple tweaking as you have to run the RC values through a filter program.
I may need some tips on the correct way to get the best results for biasing class A.
I know Steve would like me to leave it alone.
Steve will want to clone it and sell it - Haha (sorry Steve, couldn't resist :) ) but in all honesty if you have the player in a
fine state of balance you do have to proceed with caution doing any further mods that are major.
but there is still a bit to do to this player particularly things I have done in the past that i know will make it better.
One thing I did want to do is to showcase the real potential of Marantz's tda1541 , 94 mk2 and what it is capable of without changing the basic circuitry too much except for the clocking improvements .Even the NOS is not a major fundamental change to the original design .
The stock clocks are crap. So just about any after market clock will improve it to some degree.
The Valab clocks are actually not bad.
However the chip that generates the I2S signal (that feeds DAC) corrupts the clock timing quite a bit. I've got scope shots of before / after
re clocking the I2S signal and the 'before' looks pretty messy in comparison.
You can do all the usual oscon etc mods around this chip and it will make a small difference but it's still pretty messy.
The re clocker takes the I2S signals re alignes them perfectly to the clocks rising edge. It also spits out very clean wave forms.
However - it does change the sound quite a bit, so if your player is tweaked to a finely tuned balance then other things may or may not need
to be adjusted. I've tried a few and I always prefer the re clocker but it's definitely cleaner - and that's not always to everyones taste.
T
-
Yes , I was mistaken regarding the 12 v regs too . I had a look at the diagrams last night and was going to post them today . It makes life easier when these nice goodies are already there like decent discrete regs . I always find better results with nice bypasses to electrolytic caps even in good regulated supplies and even with well regarded electros .
Nice post Zen . . Some of your suggestions give me more to think about.
-
Ok , this is a photo of the servo board. Notice some silmic 2 super golds , some auricaps and other copper foil caps . Now , if its not in the amplifying section or around the dacs these areas are ignored by many . These circuits are supplied by the same power supplies as the other sensitive audio and dac circuits meaning the capacitors are basically in parallel with the other areas I spent time upgrading . Any inferior capacitor here will also affect our music making circuits . Out went the cheap electrolytics and mylar bypasses and in went the good stuff . These circuits will be noisy so top quality bypasses are important in my opinion . My guess is also that the servo could be more accurate with lower noise parts , but at the very least the power supplies have consistent top quality without weak noise generating caps . Maybe one day I will look at building a separate power supply for some voltages in there to take it off the sensitive lines .
Even worse is the ground system in these boards , which traverses from the dac board then through the decoder board then through the servo board . Noise will potentially be breaking into the system all over the place with this long and winding ground system .
I have previously separated the grounds from each board but reinstalled them because while I am modding the boards it was much quicker and easier to unplug them , but there was a clear sound improvement by using a more direct star type ground . It is sensitive to the type of wire used so need to experiment a bit .but we need probably 22 -24 g and maybe a couple of runs .
-
Hi everyone.
I also have a CD94 mkii and will be modding soon.
What I never see in reviews about the CD94 MKII is what happens when you substract the signals from 2 DAC where one had an inverted input signal.
As the clock component at 22.1 kHz is equal in both TDA-1541a and the signals are substracted the clock frequency component is eliminated and you keep the output signal 2x.
Am I right here? ;D
That is why the MKII sounds better. In theory you don't even need a filter because the clock signal is almost not there.
I will implement a lot of these mods.
Thanks for your explanation.
John
-
Hi everyone.
I also have a CD94 mkii and will be modding soon.
What I never see in reviews about the CD94 MKII is what happens when you substract the signals from 2 DAC where one had an inverted input signal.
Hi John, welcome
The two signals are opposite (one dac signal goes up +, other goes down -) and are summed in the differential to single ended circuit.
As such only signals that are the same will be cancelled. The call this common mode signals, or common mode rejection.
As the clock component at 22.1 kHz is equal in both TDA-1541a and the signals are substracted the clock frequency component is eliminated and you keep the output signal 2x.
Am I right here? ;D
Not quite.
Both DAC's outputs are stepping in opposite directions, so there is no cancellation of the 44.1kHz sample rate (if running zero oversampling).
As stated before, only signals (or noise) that is the same on both DACs will be cancelled.
That is why the MKII sounds better. In theory you don't even need a filter because the clock signal is almost not there.
This is not the case - unfortunately. As pointed out above, the 44.1kHz signal 'staircase' of zero oversampling will not
be cancelled at all.
If you still run the digital filter this will be at whatever sample rate the DF runs at, it's 4 x OS in CD94's case = 192kHz.
There's no free lunch unfortunately.
I will implement a lot of these mods.
Thanks for your explanation.
John
-
Welcome from me too John.
You should have fun modding your 94 mk2 but it is also full of frustrations at times .If you need help with anything ,let me know .
-
I will buy one of these reclockes for my stock Marantz CD92 mark 2. You cannot complain about the price.
https://www.ebay.com.au/itm/Nos-reclock-PCB-for-TDA1541A-SAA7220-based-dacs-and-players/202116316889?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2055119.m1438.l2649